S. Bell, K. Bala, N. Snavely. Intrinsic Images in the Wild. ACM Transactions of Graphics, 2014.

In contrast to the MIT Intrinsic Images dataset [1], Bell et al. introduce a much larger dataset of more realistic scenes. The ground truth was obtained using crowdsourcing. In addition, they present a new intrinsic image algorithm and include thorough comparison to several recent approaches: Garces et al. [2], Zhao et al. [3] and Shen et al. [4]. The dataset can be downloaded online at opensurfaces.cs.cornell.edu/publications/intrinsic/ and the source code is available on GitHub.

  • [1] R. Grosse, M. K. Johnson, E. H. Adelson, W. T. Freeman. Ground truth dataset and baseline evaluations for intrinsic image algorithms. International Conference on Computer Vision, 2009.
  • [2] E. Garces, A. Munoz, J. Lopez-Moreno, D. Gutierrez. Intrinsic images by clustering. Computer Graphics Forum, 2012.
  • [3] Q. Zhao, P. Tan, Q. Dai, L. Shen, E. Wu, S. Lin. A closed-form solution to retinex with nonlocal texture constraints. Transactions on Pattern Analysis and Machine Intelligence, 2012.
  • [4] L. Shen, P. Tan, S. Lin. Intrinsic image decomposition with non-local texture cues. Conference on Computer Vision and Pattern Recognition, 2008.
What is your opinion on this article? Let me know your thoughts on Twitter @davidstutz92 or LinkedIn in/davidstutz92.